10076
J. Am. Chem. SOC. 1995,117, 10076-10087
P-Hydride Elimination from Alkyl and Cycloalkyl Groups on a Cu( 100) Surface: Ring Strain and Planarity of the Transition State Andrew V. Teplyakov and Brian E. Bent* Contribution from the Depamnent of Chemistry, Columbia University, New York, New York 10027 Receiued February 13, 1 9 9 9
Abstract: Alkyl (c5-C~). cycloalkyl (C4-C7), and 2-bicyclo[2.2.1]heptyl (norbomyl) groups have been generated on a Cu(100) surface by the dissociative adsorption of their bromo derivatives. All these groups decompose by a /3-hydride elimination reaction upon heating the surface, and the product alkenes which are evolved from the surface were detected by mass spectrometry. Studies in which cyclohexyl-dl groups were generated on the surface by reacting D atoms with physisorbed cyclohexene show that the coadsorbed bromine atoms in the bromoalkane studies do not have a significant effect on the rate or mechanism of these reactions. A particularly interesting finding from these studies is that the rates of p-hydride elimination for these structurally similar alkyl groups vary by 6-7 orders of magnitude if one extrapolates the rates to a common reaction temperature of 200 K. Assuming a common, firstorder preexponential factor of lo1' s-I gives the following relative rate scale (normalized to 1 for the transformation cyclohexyl cyclohexene): cyclobutyl (0.6) < cyclohexyl(1) < norbomyl (300) < 3-hexyl (3 x IO4) % 3-pentyl (5 x IO4) < cycloheptyl = cyclopentyl (IOb), Thermodynamic differences between these reactions, as qualitatively addressed by differences in strain energy between reactant and product, may play a role in the 20-fold difference in rate between cyclopentyl and 3-pentyl, but thermodynamics cannot account for the -4 order of magnitude larger rate for 3-hexyl relative to cyclohexyl. We suggest that the dramatic rate difference between these two secondary alkyls is due to the different free energy changes required to achieve a planar geometry in the transition state, Le., a dihedral angle of 0" between the substituents on a-and B-carbons.
-
Scheme 1
1. Introduction One of the most common reaction pathways for alkyl groups bound to metal centers is p-hydride elimination. As shown in Scheme 1 for the case of a selectively deuterated ethyl group bound to a metal surface, this reaction involves, formally, the transfer of a hydrogen atom from the B-carbon of the alkyl group to the metal with concurrent formation of an alkene. From a practical standpoint, ,%hydride elimination by alkyl groups is important as (1) one of the elementary steps in the catalytic dehydrogenation of paraffins to (2) one of the pathways for chain termination in polymerization processes such as Ziegler-Natta catalysis'" and the Fischer-Tropsch (3) the rate-determining step in organometallic chemical vapor deposition processes for the deposition of metal and (4) an undersirable side reaction in catalytic olefin * T o whom correspondence should be addressed. Phone: (212) 854FAX. 1212) 972-12x9. e-mail: hent~uchem.columhia.edu.
7041.
hydrogenation" and organometallic transmetalation.".'3 The reverse of ,%hydride elimination by a metal alkyl (olefin insertion into the metal-hydrogen bond) is the important initiating step in such processes as olefin h y d r o g e n a t i ~ n , ~ ~ . ~ ~ olefin isomerization,'h olefin €IDexchange,".I7 and possibly also Ziegler-Natta catalysis.'.1x The focus of this paper is ,%hydride elimination by alkyl groups bound to a single crystal copper surface. The issue addressed is the effect of alkyl conformation on the rate of p-hydride elimination. As will be shown below, simply the difference between five-membered and six-membered carbon rings (cyclopentyl vs cyclohexyl) has a dramatic effect on the ~~~
(3) Boor. 1. J. Ziegler-Natla Catalysis and Polymerizotions; Academic Press: New York. 1979; p 670. (4) Watson, P. L. J. Am. Chem. Suc. 1982, 104, 337-339. ( 5 ) B ~ IA.. T. cad. Reu.-Sci. ~ n g 1981, . 23. 203-232. (6)Anderson, R. B. The Fischer-Tropsch Synfheryr: Academic Press. lnc.: Orlando, 1984 p 300. (7)Santilli. D. S.;Castner. D. G. Energy Fuels 1989,3,8-18. @)Bent, B. E.: Dubois, L. H.; Nuzra, R. G . Mater. Rer. Soc. Symp. prOc. 1989, 131,327-338. (9)Bent, B. E.;Nurro. R. G.:Dubois, L. H. J. Am. Chem. Soc. 1989. 1. 1 , I.h ? d. - - l.h M. .. .1 .
(10)Chan. A. W. E.; Hoffman". R. 1. Vae. Sei. Techno/., A 1991. 9. 1569-1580.
0002-7863/95/1517-10076$09.00/0
(11) Augustine. R. L. Cold. Keu.-Sci. Eng. 1976. 13.285-316. (12)Flood. T. C . In Topicr in Inorjianic and Orjianometallic Stereochemisrq; G.L. Geoffmy. G . L.. Ed.; John Wiley & Sons: New York. 1981;Vol. 12; pp 37-118. (13)Collman,J.P.;Hegedus.L.S.;Nonon,J.R.;Finke.R.G.Principle.s
and Applications of OrjinnorronritionMetal Chemirr!y; University Science Books: Mill Valley, CA. 1987;p 989. (14)Doheny. N. M.; Berkaw. 1. E. J. Am. Chem. So a
dt
m U
5 > t czn
202
P z z
0 200
300 400 500 TEMPERATURE (K)
600
7 '0
Figure 4. TPR experiments to compare cyclopentene evolution from cyclopentyl groups on the Cu(100) surface with the molecular desorption of cyclopentene: (a) monitoring m/e = 68 after a 2 langmuir exposure of cyclopentene; (b) monitoring m/e = 68 after a 2 langmuir exposure of bromocyclopentane.
displacing agent such as 1-hexene is added to bromocyclopentane monolayers which have been heated to various temperatures. If cyclopentene has been formed by heating bromocyclopentane, then it will be displaced by 1-hexene to the second layer, and in a subsequent TPR experiment cyclopentene is evolved at -140 K. If no cyclopentane has been formed, then cyclopentene is not evolved in the TPR until 200 K (the same temperature as in Figure 4). As reported in ref 62a, cyclopentene can be displaced from a Cu(100) surface by unsaturated compounds such as 1-hexene, cyclohexene, and benzene, and our studies here show that these reagents do not displace cyclopentyl groups bound to a copper surface. It is also important to note that these c6 compounds do not displace molecularly adsorbed bromocyclopentane as proven by monitoring the molecular ion of bromocyclopentane (mle = 148, 79Br isotope) after depositing 10 langmuirs of a displacing agent on a submonolayer coverage of bromocyclopentane at 100 K. As shown in Figure 5, the scheme of the experiment for determining the temperature at which cyclopentene is formed from bromocyclopentane on Cu(100) was as follows: cool the single crystal surface to at least 110 K, deposit 1 langmuir (about 25% of a monolayer) of bromocyclopentane on the surface; heat the surface to temperature T ;cool the surface back to 110 K; deposit 10 langmuirs (about two monolayers) of the displacing agent (1-hexene, cyclohexene, benzene, or iodobenzene); take a TPR spectrum tracing mass 68 (the molecular ion of cyclopentene). As shown in Figure 6 for the case where 1-hexene was used as the displacing agent, the TPR spectrum shows two peaks: one at 140 K and another at 196 K. The ratio of the areas of these peaks depends on the temperature (T) to which we heat the surface before adding the displacing agent. The lowtemperature peak (140 K) corresponds to cyclopentane which was formed during a heating to T and which was displaced from the monolayer by 1-hexene. As is evident in Figure 6, this peak first grows with increasing T and then decreases as T approaches the temperature for molecular desorption of cyclopentene from Cu( 100). The high-temperature peak (196 K) corresponds to cyclopentene which is formed from bromocyclopentane during the TPR experiment. This cyclopentene
,&Hydride Elimination fi-om Alkyl and Cycloalkyl Groups
T=128K
C-Brbond
Diemiation
I T=155K
to llOK
Displace \TPD Spectrum I ------\
I T=196K I
!a1
J. Am. Chem. Soc., Vol. 117, No. 40, 1995 10081 eV to avoid interference from the cracking pattem of I-hexene. (Nine electronvolts was the electron energy setting on the ionizer potentiometer; previous studies54have suggested that the true electron impact energy is slightly higher.) Plotting the area of the high-temperature peak versus T,we obtain a curve which decreases with increasing temperature as shown in the inset to Figure 6. Differentiation of this curve gives the rate of cyclopentene evolution from bromocyclopentane, which has a maximum at 155 K. Experiments with cyclohexene, benzene, and iodobenzene as displacing agents were performed, and the results were the same: maximum rate at 155 f 5 K. The same result is also obtained by differentiating the increasing part of the plot of the low-temperature (140 K) peak area versus T. Although the anneal/quench/displace experiments described above do not provide a highly accurate temperaturehate profile for the ,&hydride elimination reaction by cyclopentyl, if we assume that the preexponential factor for this reaction is the same as that for cyclohexyl cyclohexene conversion, then we can estimate, using the method of Redhead,60 that the activation energy for &hydride elimination in cyclopentyl groups on Cu(100) is 10.0 f 0.6 kcal/mol (heating rate taken as 0.1 Us; see Figure 9 and explanation to Table 1). The assumed value for the prefactor for cyclopentyl fi-hydride elimination translates to an experimental uncertainty of 1 kcal/ mol in activation energy assuming that the actual preexponential factor for /3-hydride elimination in cyclopentyl may differ by an order of magnitude. 3.2.c. C-Br Bond Dissociation Kinetics. The extremely low temperature for cyclopentene formation from bromocyclopentane on Cu(100) raises the question of whether C-Br bond dissociation occurs concurrently with the loss of a B-H atom in bromocyclopentane. Since the C-Br bond is difficult to detect by HREELS and R,chemical displacement was applied to determine the C-Br dissociation kinetics. A compound which displaces physisorbed bromocyclopentane from the first monolayer on Cu(100) is CH&. The fact that displacement of bromocyclopentane by CH212 is complete is based on the following evidence. For submonolayer exposures of bromocyclopentane, there is no molecular desorption in the absence of a displacing agent. However, when submonolayer exposures of bromocyclopentane are displaced by 10 langmuirs of CHZIZ, there is a linear increase in peak area with bromocyclopentane exposure and the rate of increase is the same (within 5%) as that for multilayers. The procedure for determination of the dissociation temperature for bromocyclopentane on Cu( 100) was therefore as follows: cool the surface to 110 K; deposit 1 langmuir of bromocyclopentane; heat to temperature T'; cool the surface back to 110 K; deposit 10 langmuirs of CH212; take a TPD spectrum (Figure 7) tracing mass 148 (molecular mass with 79Br isotope). The TPD spectrum shows one peak at -185 K whose area decreases as T' increases. Plotting the area of this peak versus T' and differentiating the curve, we obtain 128 K as the dissociation temperature, i.e., the maximum rate of dissociation, for bromocyclopentane on Cu(100). This result shows that dissociation does not affect the P-hydride elimination process which occurs at 27 K higher temperature. 3.2.d. Comparison of the Results of Displacement Experiments with Measurements of the Surface Work Function Change. The temperatures determined above by chemical displacement and TPR for C-Br bond dissociation (128 K), /3-hydride elimination (155 K), and cyclopentene desorption (202 K) are also apparent as inflection points in a plot of the surface work function change, A@, versus temperature. As presented in Figure 8, the work function change of a Cu(100) surface
-
Figure 5. Schematic diagram of the experimental protocol used to determine the P-hydride elimination temperature in cyclopentyl on Cu-
(100). Cu(100): 2 L bromocyclopentane heated to various temDeratures, quenched to1 lOk, and displaced by 10 L 1-Hexene
c
5
Pz 2
Q 8
$
158K
I
I
223K 200
300 400 500 TEMPERATURE (K)
-600
700
Figure 6. Determination of the ,&hydride elimination temperature in cyclopentyl on a Cu(100) surface according to the scheme in Figure 5. Two langmuirs of bromocyclopentanewas deposited on Cu(100), heated to various temperatures, and quenched by cooling to 110 K, and any product cyclopentene was displaced from the first monolayer by 10 langmuirs of 1-hexene. The TPD spectra were obtained using an electron impact ionization energy of 9 eV while monitoring d e = 68 (molecular ion). The inset shows the area of the high-temperature(196 K) peak plotted versus the annealing temperature. The inflection point
indicates the value equivalent to the peak temperature for P-hydride elimination in a TPR experiment. remains bound to the surface until the temperature of molecular desorption from the monolayer on Cu( 100) because, by the time it is formed, any second-layer 1-hexene which could potentially displace the cyclopentane as it is formed has already desorbed. It should be noted that the total area (sum of the peaks at 140 and 196 K) remains the same (within 10%) up to the temperature of molecular desorption of cyclopentene. During this experiment the electron energy of the mass spectrometer was set to 9
10082 J. Am. Chem. SOC., Vol. 117,No. 40,1995 Cu(lO0): I L bromooydopentane heated to various temperatures,' quenched to 110 K and displaced by 10 L oi CH,I, Y
Displacement
Z
annealing at:
112K 110 120 130 1 4 0 1 5 0 180
ANNEALING TEMPERATURE (K) I
123K
200
300 400 500 TEMPERATURE (K)
600
Figure 7. Determination of the C-Br bond dissociation temperature in bromocyclopentane on Cu( la)). Two langmuirs of bromocyclopentane was deposited on Cu(lOO), heated to various temperatures, and quenched by cooling the crystal to 110 K, and any undissociated bromocyclopentane then displaced from the first monolayer by 10 langmuirs of CH&. The TPD spectra were obtained monitoring d e = 148 (molecular ion for bromocyclopentane). The inset shows the area of the TPD peak plotted versus the annealing temperature to determine the temperature of C-Br bond dissociation.
150
200
250
Teplyakou and Bent experiment, the TPR peak temperatures are about 10-15 K higher than the inflection points of the corresponding parts of the A@ plot. C-Br bond dissociation and /?-hydride elimination are evident in the A@ plot in Figure 8 as an increase at 130 K and a decrease at 155 K. The fact that these inflection point temperatures agree quite well with the chemical displacement results implies that the annedquench protocol in the displacement studies is equivalent to a surface heating rate of -0.1 Ws. It is also of interest to note that the decrease in the surface work function upon bromocyclopentane adsorption and the increases in A@ upon C-Br bond dissociation and alkene desorption are consistent (in both magnitude and direction) with results in the literature for analogous pro~esses.4~ The decrease in the surface work function upon /?-hydride elimination is interesting, since, to our knowledge, this is the first measurement of the work function change for a /?-hydride elimination process which is not directly coupled to alkene desorption. 3.3. Effect of Coadsorbed Bromine. The presence of bromine atoms on the surface does not significantly affect the @-hydrideelimination temperature as evidenced by experiments with atomic deuterium. When reacted with preadsorbed cyclohexene on the Cu( 100) surface, atomic deuterium attacks the double bond63 and forms a cyclohexyl group bound to the surface analogous to the species formed by dissociation of bromocyclohexane. When this cyclohexyl-dl group undergoes @-hydrideelimination,the product [deuterated cyclohexene ( d e = 83)] can be detected by TPR. As seen from Figure 3c, this mass shows a distinctive three-peak spectrum. The peak at 21 1 K corresponds to the A 1 peak of unreacted cyclohexene (due to the 1% natural abundance of I3C; compare Figure 3a and note the expansion factor); the second peak (at 239 K) has the same temperature as the main peak in the experiment with bromocyclohexane (Figure 3b) and corresponds to cyclohexenedl evolution after /?-hydride elimination by cyclohexyl-dl groups;63 the peak at 280 K is due to the desorption of cyclohexene and cyclohexene-dl from surface defect sites. The fact that the 239 K peak for cyclohexene-dl in Figure 3c in the absence of surface bromine matches the 239 K peak for cyclohexene evolution from bromocyclohexane where bromine is coadsorbed with cyclohexyl groups indicates that coadsorbed bromine does not have a significant effect on the rate of cyclohexyl /3-hydride elimination on Cu( 100). Similar results for ethyl groups on Cu( 111) in the presence and absence of coadsorbed iodine are presented in ref 63.
+
300
TEMPERATURE (K)
Figure 8. Comparison of the surface work function change with data obtained by conventional TPD experiments (see Figure 4) and displacement methods (see Figures 6 and 7). The work function change was measured after deposition of 2 langmuirs of bromocyclopentane on Cu(100). The heating rate in these studies was 0.1 IUS.
covered with 2 langmuirs of bromocyclopentane shows four features. The two highest temperature features correspond to molecular desorption of cyclopentene from the (100) face and from defect sites, respectively. Since the heating rate in these studies was only 0.1 Ws, compared with 3 Ws in the TPR
3.4. Bromocyclobutane, exo-2-Bromobicyclo[2.2.l]heptane (exo-2-Bromonorbomane), 3-Bromohexane, 3-Bromopentane, and Bromocycloheptaneon Cu(100). These compounds, like bromocyclopentane (section 3.2) and bromocyclohexane (section 3. I), undergo C-Br bond dissociation followed by /?-hydride elimination on Cu(100). The main differences among these compounds are in the relative surface reaction rates which were determined using the methods described above for bromocyclopentane and bromocyclohexane. These various alkyl groups were chosen because they represent a spectrum of properties such as ring strain and relative conformational stabilities. At the same time, the electronic stuctures of all these secondary alkyl groups are similar. The results are summarized in Figure 9, which compares TPR spectra for alkene evolution from the @-hydrideelimination reaction with TPD spectra for molecular desorption of the product alkene from Cu(100). The interpretation of these results is given briefly below by compound. 3.4.a. Bromocyclobutane. This compound was studied to investigate the effect of ring strain in cyclic alkyls on the rate of @-hydrideelimination. Cyclobutane has 26.5 kcdmol of (63)Xi,M.; Bent, B. J . Vac. Sci. Technoi., B 1992,10, 2440-2446.
P-Hydride Elimination from Alkyl and Cycloalkyl Groups
J. Am. Chem. SOC.,Vol. 117, No. 40,1995 10083
m
18 K higher in temperature than the temperature of molecular desorption of norbomylene (see adjacent panel in Figure 9). The 270 K peak is due to rate-determining product desorption 'f2 from surface defect sites (see section 3.1 for the explanation). exo-2-Bromonorbornane As in the case of the cyclohexyl cyclohexene reaction on Cu( loo), the rate of norbomylene evolution from the surface 2 r '1' BromocvcloheDtane C ycloheptene corresponds to the rate of P-hydride elimination, and the :/e=96 Q activation energy for this conversion, as determined by the 1.Hexene heating rate variation method, is 16.9 ic 0.6 kcal/mol. The m/e=84 corresponding prefactor is an unexpectedly large (4.7 f 4.2) x 3-Bromopentane 2-Pentane l O I 7 S - I . ~ It is possible that the coverage independence of the kinetic parameters which is assumed in this method is not valid 200 300 400 500 200 300 400 500 for this reaction, and this value should be treated with appropriTEMPERATURE (K) TEMPERATURE(K) ate caution. Figure 9. Comparison of TPR results for alkene evolution from 3.4.c. Bromocycloheptane on Cu(100). Cycloheptyl is of P-hydride elimination [after a 2 langmuir exposure of the indicated particular interest for comparison with cyclopentyl. Not only bromoalkanes on Cu(lOO)] with molecular desorption studies of 2 is the increase in strain energy on converting the C5 and C7 langmuirs of the alkenes. Because of availability, 1-hexene was used alkanes to the corresponding alkenes quite similar$6 but also instead of 2- or 3-hexene to compare with P-hydride elimination by the conformational energy differences in these cyclic alkanes 3-hexyl; cyclobutene molecular desorption was not studied. are quite similar to one another and much smaller than those in cyclohexane. Not surprisingly, the rate of P-hydride elimination strain energy compared with 6.2 for cyclopentane and 0.0 for from cycloheptyl on Cu(100) is similar to that for cyclopentyl cyclohexane6! and orders of magnitude faster than that for cyclohexyl (see As for bromocyclohexane, P-hydride elimination in cyclobutyl below). In reaching this conclusion, it is assumed that, as for occurs above the temperature for cyclobutene desorption from bromocyclopentane, C-Br bond dissociation in bromoheptane Cu( loo), so the @-hydrideelimination kinetics can be determined occurs at temperatures below 160 K and consequently does not directly from the cyclobutene TPR spectrum (monitoring mass affect the temperature of P-hydride elimination. 54) as shown in Figure 9. This conclusion presumes that, by Chemical displacement was used to determine the temperature the 244 K peak temperature for cyclobutene evolution, the C-Br of P-hydride elimination in cycloheptyl since the temperature bond has already dissociated and that the temperature of of cycloheptene evolution from cycloheptyl is the same as the molecular desorption of cyclobutene lies below 220 K. The temperature of the molecular desorption of cycloheptene. A first assumption is consistent with the measured C-Br bond . ~ ~ ~ ~ useful displacement agent for this situation was found to be dissociation below 200 K for other b r o m ~ a l k a n e s $ ~and o-xylene. However, this displacing agent desorbs from the the second assumption is reasonable on the basis of studies of second layer at -180 K on Cu( loo), which means that it remains the molecular desorption of various hydrocarbons from the Cuon the surface as a second layer long enough to displace the (100) surface: 1-butene desorbs molecuarly at about 170 K, alkene as soon as it is formed by @-hydrideelimination in the and cyclobutene would be expected to desorb at a similar first layer. As a result, o-xylene displaces cycloheptene to the temperature since there is no significant difference in the second layer as it is formed by @-hydride elimination from temperature of molecular desorption between 2-pentene and cycloheptyl. The /?-hydrideelimination temperature determined cyclopentene on Cu( 100). The possibility of isomerization of by this method is -155 K. It should be noted, however, that the product cyclobutene to 1,3-butadiene appears unlikely on the basis of a comparison of the mass 54 TPR spectrum for the high coverages of displacing agents present during the bromocyclobutane with the TF'D results for molecular desorption P-hydride elimination reaction in these studies may have had some effect on the reaction kinetics as a result of, for example, of 1,3-butadiene. Both spectra show peaks near 240 K, but the site-blocking. For all other systems studied (for reasons which substantial tailing to higher temperature for 1,3-butadiene argues are not yet understood) this approach of immediate displacement against a common reaction product. by a physisorbed second-layer molecule did not work; the The activation energy for P-hydride elimination by cyclobutyl product alkene could not be displaced from the first monolayer on Cu(100) was determined by heating rate variation studies as it was formed, even if the displacing agent did displace the (see section 3.1) to be 16.3 f 0.5 kcallmol. The Arrhenius alkene in control experiments and had a large enough heat of factor corresponding to this energy is (3.0 f 2.3) x l O I 4 s - I . ~ O adsorption to remain on the surface as a second layer until the 3.4.b. exo-2-Bromonorbornane. This compound is of temperature of the P-hydride elimination. interest to compare with bromocyclopentane. Similar to bromocyclopentane,the substituents on the 2- and 3-carbon centers 3.4.d. 3-Bromohexane and 3-Bromopentane. Studies of (Le., the Br-substituted carbon and the adjacent CH2 group) of 3-pentyl and 3-hexyl provide an opportunity to compare norbomyl are nearly eclipsed, and this group would thus be P-hydride elimination in acyclic compounds with their cyclic expected to undergo @-hydrideelimination at low temperature counterparts. As shown in Figure 9, P-hydride elimination in from the point of view of a planar transition state. On the other these alkyls occurs at or below the temperature of molecular hand, unlike for the C5 system, one expects a significant increase desorption of the product alkenes. in ring strain in transforming norbomyl to norbomylene (4.8 Studies of 3-bromopentane show that the temperature of kcaYmol for norbomane norbomylene in the gas phase4% C-Br bond dissociation in this acyclic bromoalkane is 128 K The results in Figure 9 show that, after 4 langmuirs of exo(the same as for bromocyclopentane) as was found in experi2-bromonorbomane was deposited on the Cu( 100) surface, the ments analogous to that described in section 3.2.c for bromocyd e = 94 (molecular mass of the product norbomylene (bicycloclopentane with CH212 as a displacing agent. The temperature [2.2.l]heptene)) TPR spectrum has two peaks: a primary peak of P-hydride elimination by the 3-pentyl groups generated on a at 210 K and a smaller peak at 270 K. The 210 K peak is surface by C-Br bond dissociation was determined by chemical displacement. o-Xylene was used as the displacing agent, and (64)Lin, J.-L.; Bent, B. E. Chem. Phys. Lett. 1992,194, 208-212. the displacing ability of this compound was verified by (65) Lin, J.-L. Dissertation, Columbia University, 1993.
1
244
Bromocyclobutane
I
I
-
-
10084 J. Am. Chem. SOC., Vol. 117,No. 40, 1995
Teplyakov and Bent
Table 1. &Hydride Elimination Temperatures and Corresponding Kinetic Parameters for the Indicated Alkyl Groups on Cu( 100)
alkyl group
d'
0'
P-H elimination temperature: K
k,b s-I
244 155'
0.091 -0.1
239 155e
0.065 -0.1
210
E, (Redhead method) v = lOI3 v = lOI5
E,
(heating rate var)
v = 10"
15.3
16.3 f 0.5
12.6
14.9
17.1
7.9
9.3
10.7
-106
12.3
14.5
16.7
1
7.9
9.3
10.7
-106
10.8
12.7
14.7
9.6 15.1
15.5 f 0.5
9.6
0.191
12.8
30.1
16.9 f 0.6
I
&
relative rate at 200 Kd
AG*: kcal/mol
-0.6
-300
175e
A
10.8
8.9
10.5
12.1
?. 4 x
177'
-0.1
11.0
9.0
10.6
12.2
-3 x 104
104
a Peak temperature from TPR experiments; the surface heating rate is 3 Ws. First-order rate constant at the temperature given in column 2; see section 3.5 for a discussion of the method of calculation. On the basis of an estimated absolute accuracy of f 1 0 K in the surface temperature measurements and an assumed coverage and temperature independence of the kinetic parameters, we estimated that these parameters are accurate to within at least an order of magnitude. The values associated with the temperatures indicated by footnote e have slightly higher uncertainties due to the experimental method (see footnote e ) . Uncertainty based on 1 order of magnitude uncertainty in the rate constant (see footnote b ) is about f1 kcaymol. Common first-order preexponential factor of loT3s-I used. E Temperature determined by chemical displacement and/or work function change measurements (these temperatures are approximately equivalent to what would have been determined in TPR experiments with heating rates of -0.1 Ws).
displacement experiments in which 2-pentene was adsorbed and displaced by o-xylene. The identity of the product (distinguishing 1-pentene and 2-pentene is possible by mass spectrometry) was determined by monitoring characteristic peaks in its mass spectrum. The results show that the temperature of /?-hydride elimination is 175 K. A procedure directly analogous to that described above was also followed to determine the /?-hydride elimination temperature for 3-hexyl. o-Xylene (or iodobenzene) was used as displacing agent, and the temperature of /?-hydride elimination for 3-hexyl was found to be 177 K.
3.5. Reaction Rate Constants and Free Energies of Activation. TPDmPR experiments are generally not sufficient to determine accurately (or at least unambiguously) the activation energies and preexponential factors for surface reactions.57 The simultaneous changes in coverage and temperature in these experiments coupled with the often significant coverage dependences of the kinetic parameters do not allow a unique determination of the activation energy and prefactor. However, if the kinetic order for reaction is known, the surface reaction rate constant at any specified temperature where there is a measurable signal in a TPDRPR experiment can be determined to within the accuracy to which the absolute surface coverage is known. While absolute surface coverages are often difficult to determine with accuracy at the percent level, even accuracy to within a factor of 2 (which is not difficult to obtain) is sufficient, as we show below, to determine free energies of activation to within kT of the accurate value, Le. within 1 kcall mol for a temperature of 300 K. For example in the case of a first-order reaction (which is a reasonable approximation for the rate of /?-hydride elimination by alkyl groups at low coverages on Cu(lOO), the reaction rate can be expressed as rate = -(dO/dT)B = k,8 where kl is the first-order rate constant (s-l), 8 is the absolute surface coverage of reactants (molecules/cm2), T is the surface temperature (K), and fi is the surface heating rate (Ws). To determine the reaction rate constant at any specified temperature, one first determines the proportionality constant relating the TPD
signal to the absolute rate by equating the absolute initial surface coverage to the TPD peak area (integrated vs time). The rate constant for any particular temperature, T, is then determined from the relation
k , ( T ) = rate(T)/B(T) In the /?-hydride elimination studies here, the rate constants at the TPR peak temperatures have been determined assuming first-order kinetics and using the procedure described above. (Given the possibility of a non-fist-order process, this assumption introduces an experimental uncertainty of up to a factor of 2 for the coverages at the TPR/D peak maximum.) The absolute surface coverages have been estimated by equating the monolayer saturation coverage with the maximum surface density possible based on the van der Waals size for the alkyl groups and bromine atoms which were packed on the surface in orientations which maximize dispersion interactions with the surface. On the basis of the literature data, this procedure can be expected to give results which are accurate to within about a factor of 2, which when combined with the uncertainty in the reaction order gives an accuracy to within a factor of 4. The results of these rate constant determinations are summarized in Table 1. Note that all rate constants are quite similar at the TPR peak temperatures. This is to be expected. For a given heating rate, reactions which have the same reaction order and prefactor will have similar peak height to area ratios, Le., similar peak shapes.& Redhead has shown that the ratio of these proportionality constants is almost the same for all first-order reactions if the ratio is evaluated at the same extent of the reaction, for example, at the peak temperature.@' Thus, for fistorder reactions with a common preexponential factor, the rate constant, as determined by eq 2 above, will always be -0.1 s-' when evaluated at the various TPR peak maxima. It should be emphasized, however, that the different reactions will have very different rate constants if a common reaction temperature is chosen. The second point to note in connection with Table 1 is that, for the values associated with temperatures indicated by footnote (66)Yang, M. X. Manuscript in preparation.
J. Am. Chem. SOC., Vol. 117, No. 40, 1995 10085
/%Hydride Elimination from Alkyl and Cycloalkyl Groups e, the uncertainty is larger than a factor of 4 because the surface heating rate is not accurately known. Specifically, for these alkyl groups, the rate of /%hydride elimination was monitored by chemical displacement or work function change methods where the surface reaction was repeatedly initiated (by heating) and quenched (by cooling). The studies in connection with Figure 9 suggest that this initiatelquench procedure is equivalent to a surface heating rate of 0.1 Ws, but we estimate that an uncertainty of up to a factor of 2-3 is introduced by this approximation. When combined with the uncertainty in surface coverage and reaction order, this gives a total uncertainty of a factor of 10 in the rate constant. Within the framework of transition state theory, the free energies of activation can be uniquely determined from the rate constants in Table 1 using the following relations: k,(T,) = kT,/h exp(-A&/RT,) A& = -RTp ln(k,h/kT,)
(3)
(4)
where k is Boltzmann’s constant, h is Planck‘s constant, and Tp is the TPR peak temperature. The uncertainties in AG* are RT, In a where a is the uncertainty in the rate constant. Thus, an uncertainty of a factor of 10 in the rate constant translates to 1 kcdmol for these P-hydride elimination reactions which occur at 150-250 K.
-
4. Discussion Table 1 summarizes the kinetic parameters that were determined from the thermal desorption, chemical displacement, and work function change studies described above. While there is some uncertainty in the kinetic parameters (particularly the activation energies and preexponential factors), all of the standard methods of analysis show that there is a dramatic variation in the rates of P-hydride elimination across this series of alkyl groups on Cu(100). For example, if one extrapolates the rates to a common temperature of 200 K using activation energies determined assuming a common preexponential factor 0f10’~ s-’, then one finds (see last column of Table 1) that the rates vary by about 7 orders of magnitude. These dramatic rate differences are not unexpected on the basis of what is known about elimination reactions. As mentioned in the Introduction, there are numerous examples where 1,2-eliminations from C5 rings are orders of magnitude faster than the corresponding eliminations from c6 ring^.^^,^^ This result is analogous to what is found here for 8-hydride elimination for cyclopentyl vs cyclohexyl. Also, Kemba1167and Burwe116*have reported that catalytic H/D exchange around one face of cyclopentane on metal catalysts can be much faster than WD exchange around one face of cyclohexane. Since this process presumably involves alkyvolefin equilibration on the surface via P-hydride elimination, it is not surprising that 8-hydride elimination is significantly faster in a (25 vs c(5alkyls. 4.1. Rationalizing Relative Rates: Entropy vs Enthalpy and Possible Thermodynamic Effects. We tum now to the reason for the dramatic rate variations between these structurally similar alkyls-all of which are bound to the surface through a secondary carbon atom and all of which have single alkyl substituents on both the a-and @-carbons. No doubt, differences in both the enthalpy and entropy of activation play some role; however, to account for the dramatic effects observed, enthalpy effects must be dominant. Minimum and maximum entropies of activation expected for a cyclic four-center transition state like that expected for P-hydride elimination are -4 and 0 eu, (67) Kemball, C. Adu. Card. 1959, 11, (68) Burwell, R. L., Jr.; Shim, B. K. C.; Rowlinson, H. C. J . Am. Chem. SOC. 1957, 79, 5142-5148.
r e s p e ~ t i v e l y . ~While ~ these limits are for the gas phase elimination reactions, freezing additional rotations in a surface process would be expected to contribute at most 8 eu. Using this upper limit, the rate constant (and rate) varies by -1.5 orders of magnitude, which is much less than the 7 orders of magnitude observed. In addressing why the enthalpies of activation might be different in these systems, it is useful to distinguish between changes in the enthalpies of reaction (which would also be expected to show up to some extent as changes in the enthalpies of activation) and changes in the enthalpy of activation without a corresponding change in the reaction thermodynamics. Unfortunately, AH for these P-hydride elimination reactions is unknown. AH for /3-hydride elimination by ethyl groups on Cu( 100) has been determined to be 6.5 f 4 kcal/mol,7° but the procedure used to derive this value (i.e., measuring the activation energy for the reaction in the forward and reverse directions) was not possible for the systems described here where no significant reverse reaction was detectable. It should be noted, however, that the lack of reverse reaction simplifies the kinetic analysis, since one does not have to worry about the enthalpy change from an alkyvalkene preequilibrium contributing to the activation energy for reaction. In other words, the absence of any preequilibrium has already been assumed in Table 1 by equating the activation energies measured for 8-hydride elimination coupled with alkene desorption to the activation energies of P-hydride elimination. 4.2. An Assessment of A M for /?-Hydride Elimination on Surfaces. While the surface reaction thermodynamics for these various systems are unknown, we believe that a reasonable estimate of differences between the surface reaction enthalpies, i.e., AAH, can be made on the basis of the gas phase values of AAH for the dehydrogenation of alkanes to alkenes. The basis for this comparison is the formal analogy given in Scheme 2
Scheme 2
M
H
where M refers to the metal surface. Clearly the enthalpy changes for the two systems in Scheme 2 are quite different, but one might expect similarities among the AAH values for a series of structurally similar alkanedalkenes in these two systems if it is differences in the hydrocarbon portion of the system which dominate AAH. In the gas phase there is indeed evidence that this is the case. If one replaces an H in the alkane (at the site adjacent to the incipient n-bond) with a CH3 group to form a methylated alkane, one finds that the A M values for loss of CH3-H from the methylated alkane are quite similar to those for loss of H-H from the corresponding alkane. Whether or not the same is true for loss of M-H has not been demonstrated, but it is reasonable to assume that steric and electronic effects of the metal (while different for the reactant and product) would be similar among the structurally-related hydrocarbon systems studied here and therefore while affecting AH would not contribute significantly to A M . Since AAH values for alkane dehydrogenation are essentially equivalent to what are known as “olefin strain energies”, it is the commonly tabulated olefin strain energies which are given (69) O’Neal, H. E.; Benson, S. W. J . Phys. Chem. 1967,71, 29032921. (70) Jenks, C. J.; Xi, M.; Yang, M. X.; Bent, B. E. J . Phys. Chem. 1994, 98, 2152-2157.
Teplyakov and Bent
10086 J. Am. Chem. SOC., Vol. 117, No. 40, 1995 Table 2. Comparison of AAQ for /%Hydride Elimination by Surface Alkyl Groups (Relative to Cyclopentyl) with Olefin Strain Energy and the Minimum Energy (AE) Required To Achieve Planarity between the Substituents on a-and /%Carbons for the Indicated Alkane- Alkene Pairs in the Gas Phase alkane
alkene
0
Q
0 0
olefin strain AE,b AAG*B-H, energy,a kcal/mol kcal/mol kcal/mol -2.1
-0
0
-2.1
-0
0
(3.4‘) -1.1
4.8 -0.3 1.9
3.5
-0
1.2
3.2
5.6
5.5
1.4
5.1
“ I n the case of acyclic alkyls, we have used AAH of olefin dehydr~genation~~ (normalized to cyclohexene olefinic strain equal -0.3 kcaUmo1) as a measure of the olefin strain energy (see text section 4.2). Values given are for trans isomers; cis isomers will be -0.5 kcaU mol less negative. AE is the difference in energy between the most stable alkane conformation and the lowest energy alkane conformation (not necessarily stable) having eclipsed interactions between C-H bonds on adjacent carbons. Two values are given for cycloheptane because AE in this system depends on which pair of carbon atoms we focus; Le., depending on the conformation and the point of attachment of cycloheptyl to the surface, one expects either an existing adjacent eclipsed interaction or a significant energy bamer to achieve one.
in Table 2. Before discussing the implications of these olefin strain energies for ,&hydride elimination on Cu( loo), we comment briefly on the meaning of olefin strain. Olefin strain energy is a measure of the differences in the amount of “strain” inherent in the alkene and corresponding alkane. This strain arises from a complex combination of factors such as bond angle distortion, bond extension or compression, torsional angle distortions, and nonbonded repulsion^.^^ It should be emphasized that while the hydrocarbons listed in Table 2 have a wide range of strain, the important aspect with respect to the reaction thermodynamics is the difference in the amount of strain between reactant and product, and this quantity is the olefin strain energy. 4.3. Possible Correlation between Olefin Strain Energy and AAG* for @-HydrideElimination. Using olefin strain energy (OSE)46 as a measure of AAH for /3-hydride elimination among the alkyls in Table 2, we see that many of the trends in A A b can be rationalized. In other words, there appears to be some correlation between changes in the reaction thermodynamics (as qualitatively assessed by OSE) and changes in the reaction kinetics (as quantitatively measured by AAG*). For example, cyclopentene and cycloheptene have similar amounts of olefin strain energy, and the corresponding alkyls have virtually identical free energies of activation. Cyclobutene with about 4 kcallmol more olefin strain energy has a free energy of activation which is -5 kcallmol larger. However, one clue that changes in the reaction thermodynamics cannot completely account for the differences in rate is the fact that one generally expects only part of AAH for reaction to show up as a change in activation energy. For example, in
the empirical Ogg-Polanyi relationship7’for structurally analogous reactions
E, = E,’ = a(AH - AH’) the value of a is generally between 0 and 1. On the other hand, if one takes the AAblOSE correlation literally, then a is in some cases even greater than 1 for the systems given in Table 2. For example, the OSE ( A m for cyclohexene is 1.8 kcall mol larger than that for cyclopentene, yet AAG*(A&) for the corresponding alkyl p-elimination reactions differs by 5.5 kcall mol. The reverse situation exists for norbomyl. A@ is much smaller than expected on the basis of the OSE. In the case of norbomyl, we suspect that the correlation may fail because of the extremely rigid conformational constraints and the large steric bulk of this alkyl group. It is probable that the amount of steric hindrance between the alkyl and the surface is significantly larger than for the other alkyls, while similar differences are probably not operative for the product alkene which is only weakly coordinated to the surface;56the net effect is that the reactant alkyl may be destabilized relative to the other alkyl groups, thereby giving rise to the significantly lower A& than one would expect on the basis of the increase in strain during the reaction. On the other hand, the structural similarities among cyclohexyl, cycloheptyl, and cyclopentyl suggest that factors other than sterics are necessary to account for the lack of correlation between OSE and AA@. The predominant effect, we believe, is relative conformational stabilities and the necessity of planarity in the transition state as discussed below.
4.4. Evidence for a Planar Transition State for /?-Hydride Elimination. One possible kinetic explanation for the dramatic effect of alkyl structure on the /3-elimination rate is the inductive effect of substituents on the a- and /?-carbon atoms. It has been shown in the literature** that inductive electronic effects can have order of magnitude effects on the rates of ,&hydride elimination. In the studies here, however, all of the alkyls studied are bonded to the surface through a secondary carbon and each has a single alkyl substituent at both the a- and /?-positions. One would thus expect similar inductive electronic effects. Furthermore, one would also expect these structurally similar alkyls to have similar steric interactions with the surface. A more likely explanation is that alkyl conformation plays an important role. The relative conformational energies in cyclic and acyclic compounds vary significantly, and as discussed in the Introduction, both theory and experiment suggest that a planar transition state (i.e., a specific conformation) is required in concerted 1,2-elimination reactions which convert hydrocarbons to olefins. Conceptually, a planar transition state can be rationalized as necessary in order to maximize the overlap between the p-orbitals in the incipient n-bond in these elimination reactions. The results here are consistent with the idea that a similar conformational geometry is also important for surface &hydride elimination reactions. The strongest experimental evidence is the dramatic differences in rate for cyclohexyl vs cyclopentyl and cycloheptyl-differences which do not appear to correlate with changes in the surface reaction thermodynamics. These differences can be rationalized on the basis of the very different energies required to achieve a conformation where the M-aC and PC-H bonds lie in the same plane. While these energies are not known on the surface, the gas phase values for achieving an eclipsed conformation of adjacent C-H bonds in the _
_
_
_
~
~
_
_
_
_
_
_
~
~
~
~
(71) This relation was firstly published by M. G. Evans and M. Polanyi (Trans. Faraday SOC.1938,34, 11-29) but the “Ogg-Polanyi” descriptor was used by J. C. Polanyi (Chem. Phys. Lett. 1967, I , 421-423). (72) Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermodynamical Data oforganic Compounds; 2nd ed.; Chapman and Hall: London, 1986; p 792.
8-Hydride Elimination from Alkyl and Cycloalkyl Groups
Figure 10. Acceasihility 01the planar transition state far P-hydride elimination in cyclopentyl and cyclohexyl groups on a surface: (a) cyclopentyl (envelope conformation):(b) cyclohexyl (chair conformation); (c) cyclohexyl (boat conformation),
corresponding alkanes are given in Table 2. It is clear from these values that, unless things change dramatically on the surface, achieving an eclipsed conformation in cyclohexyl requires significantly more energy than for the CSand C7 cyclic alkyls. In fact, if the 5.5 kcallmol difference for the gas phase were to show up in AA@ for the surface ,!?-elimination reactions. then the rate for cyclohexyl would be a factor of IO6 smaller than that for cyclopentyl. which is close to what is observed experimentally. Schematic representations of the relevant structures are shown in Figure IO. Similarly, for the other alkyls (except norbomyl, which we believe to be an exception as discussed above), the values for achieving an
J. Am. Chem. Soc., Vol. 117, No. 40, 1995 10087 eclipsed conformation between adjacent C-H bonds are consistent with the relative rates observed. One final note: On the surface, defining a planar transition state in terms of the M-'LC bond can be ambiguous depending on the alkyl adsorption site and the points on the surface to which one assumes that the alkyl group is bonded. A rigorous definition of a transition state geometry which maximizes p-orbital overlap in the incipient n-bond is that the dihedral angle between the substituents on the a- and ,!?-carbon atoms should be zero. 5. Conclusions The results here show that a series of secondary alkyl bromides react thermally on a Cu(100) surface by formal 1,Zelimination of HBr to evolve the corresponding alkene at temperatures below 250 K. These elimination reactions occur by a two-step process involving first C-Br bond dissociation and subsequently ,!?-hydrideelimination. The H and Br atoms produced by these reactions remain adsorbed on the Cu(100) surface up to temperatures of 300 and 930 K. respectively. While the rates of C-Br scission are quite similar among these alkyls, the rates of ,!?-hydride elimination (when extrapolated to a common temperature) vary by orders of magnitude. Comparison of the measured free energies of activation with qualitative assessments of the reaction thermodynamics and with the expected conformational requirements for the reaction provides evidence for the important role of ring strain and planarity in the transition state.
Acknowledgment. Financial support from the National Science Foundation (Grant CHE 93-18625) and from the Camille and Henry Dreyfus Foundation as part of their TeacherScholar Program (B.E.B.) is gratefully acknowledged. We thank R. McAllister for a loan of the Kelvin probe and M. X. Yang for setting up the probe to make work function change measurements. JA950497Q
